This study compared the results obtained with two measures of alcohol consumption, a quantity-frequency (QF) method and a retrospective diary method, using a worksite sample. In order to control for the possibility of order effects, a subsample was chosen for readministration of the alcohol measures, when the order of presentation of the two measures was reversed for half of the subsample. Varying the order of presentation of the alcohol consumption items did not significantly affect the results obtained, except for an interaction effect between drinking group and order of presentation. Although subjects reported greater alcohol use on the diary than on the QF method at lower levels of consumption, the findings were reversed for higher levels of consumption. The relationship between the two measures is linear, with slope less than 1. Overall, subjects reported statistically significantly more alcohol consumption on the diary (mean [+/- SD] = 18.3 +/- 19.1 drinks) than on the QF method (mean = 16.9 +/- 19.5 drinks). The diary measure identified 20.9% of drinkers reporting consumption of more than 28 drinks per week, compared with 17.8% for the QF. However, the diary measure identified only 10.6% of drinkers reporting more than 42 drinks per week, compared with 10.2% for the QF. Consequently, the usefulness of the diary in detecting heavy drinkers appears to decrease with increasing consumption. These findings suggest the need for careful selection of alcohol consumption measures for research purposes and the need for caution in comparing the results of studies using different measures of alcohol consumption.